Based on previous research on super-resolution and colour correction by example, the present project employs such techniques towards a high-quality reclamation of lost art. The aim is to produce high-quality images of nowadays destroyed or missing paintings using the correspondence between similar artworks available in both low and high quality images. Several approaches for both super-resolution and colour correction techniques have been studied, implemented and tested to result to the most efficient and appropriate to be used. This project is an attempt that has never been done before and the successful reclamation of lost art initialises a new area of colour imaging applications in fine art. It reveals unlimited possibilities in the domain and establishes the potential of further attempts in this direction.iii Computer-Aided Reclamation of Lost Art SummaryThe present document covers the entirety of a proposed approach and an implemented solution towards Reclamation of Lost Art. Employing a combination of example-based super-resolution and colour correction techniques, the project attempts a unique step in colour imaging applications in fine art. It explores the applicability and performance of state-of-the-art techniques in super-resolution, inverse halftoning, image registration, colour transfer and image blending methods to the demanding image set of fine art specimens.Firstly, in Chapter 1, the problem at hand is introduced and stated along with the motivation behind this aim. It expands on the significance of contribution of this endeavour, the implicated considerations and relevant previous research.

Implementations of Example-Based Super-Resolution (EB-SR) have been developed extensively. Any such EB-SR method is typically evaluated against a constructed test set as to define its performance and applicability. Nevertheless, it is rare for a formed test set to precisely resemble data met in a real-world problem. Usually, low-quality training and test subsets are obtained directly from their corresponding high-quality ground truth data. This allows for a complete and reliable quantitative examination of performance at a later stage. In a real-world problem however, test data are obtained from another source, as for example, printed images. Naturally, low-quality scanned halftones and high-quality continuous tone images would possibly be spatially incoherent training pairs. Such circumstances give rise to one major consideration, misalignment in training subsets. The present work demonstrates the significance of effect of misalignment among training subsets in applying EB-SR and supports the necessity of image registration in preprocessing to overcome this problem.